Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

An ASI1600 high gain experiment - Bubble Nebula

This topic has been archived. This means that you cannot reply to this topic.
120 replies to this topic

#1 Jon Rista

Jon Rista

    ISS

  • *****
  • topic starter
  • Posts: 24,340
  • Joined: 10 Jan 2014

Posted 04 September 2016 - 03:21 PM

Last night ended up clearing rather suddenly about a quarter past 9pm. I hadn't planned on it being clear, and I wasn't sure if it was going to stay clear...but when the skies remained spotless by 9:30pm, I put out my gear. I had been imaging Sh2-132 and Pacman nebulas, however I promised many of you ASI1600 fans a while back that I would so some high gain NB imaging with short exposures to see how it went. When I set up, Sh2-132 was a bit far along, and Pacman wasn't quite clear of the trees yet...so I zeroed in on Bubble Nebula, about midway between the two. I usually use a high gain setting of 300 to do my framing and focusing, as with relatively short exposures you can still see the objects you are imaging, and with 1s subs you can focus the stars in real time. Rather handy.

 

Well, I noticed how clean Bubble looked with only 90 second test subs when I was framing...and decided to embark on a little high gain, short exposure narrow band project. I stuck with the high gain, although backed it off to setting 200, and stuck with the 90 second subs. I decided to acquire 16 Ha (3nm AstroDon, 150mm f/4 refractor) from my red zone, which last night looked like it was around 18.9mag/sq". While the first 16 subs were exposing (my original plan was to switch to Pacman OIII once it cleared the trees), I fiddled around with the first sub:

 

0fNyAoi.jpg

 

Not bad at all for 90 seconds, grand total integration, for a 3nm narrow band filter! I decided to forego Pacman, and get some SII and OIII on Bubble at the same gain setting and with the same exposure lengths. The project turned out to be quite a success. I acquired 36 subs in all three channels, 54 minutes each. The full Ha stack, 46.5 minutes worth (31 good subs) turned out very nice:

 

9uFhgxk.jpg

 

I decided to try and balance the SNRs of each channel, so I stacked 9x90s Ha, 25x90s OIII, and 36x90s SII...progressing from highest SNR per sub to lowest. The SII here was quite faint...a good deal fainter than the OIII, but definitely still there. I decided to first generate a master luminance from all three channels after an initial blend:

 

niDJOMl.jpg

 

The SII really brings in a lot more stars than either Ha or OIII. Both contain the same stars, but not nearly as intense as in SII. I've noticed that before when working other NB images...but the difference was quite stark this time. For 1h45m grand total of integration, I think the experiment was a success! I am currently processing an SHO custom blend version, and will share the results once I have a final version. Until then...if you are stuck in the city and have limited time...consider high gain imaging with a modern CMOS camera from ZWO or QHY. I acquired a full set of data across all three primary NB channels for Bubble and the field surrounding it in a matter of 4 hours active imaging time (and only 1h45m worth of integrated data)! Never did I imagine that I'd be able to create a full color tri-channel NB image with so little time invested...however things are changing. New technology is really opening up some doors. 



#2 Jon Rista

Jon Rista

    ISS

  • *****
  • topic starter
  • Posts: 24,340
  • Joined: 10 Jan 2014

Posted 04 September 2016 - 03:56 PM

Alright. After a custom SHO blend in linear ode, some TGV and MMT, a stretch and some color processing, here is what 9x90s Ha (13.5 minutes), 25x90s OIII (37.5 minutes) and 36x90s SII (54 minutes), a total of 1h45m, looks like:

 

Lukz863.jpg

 

Even with only 9 subs, the Ha here is quite overpowering. I had to make some compromises with color to get a decent blend with it. I think about double the integration time in OIII and SII would help balance that out, however, and deliver a stellar image. An HOO blend is next. 


Edited by Jon Rista, 04 September 2016 - 03:57 PM.


#3 shawnhar

shawnhar

    Cosmos

  • *****
  • Posts: 9,452
  • Joined: 25 Jun 2010

Posted 04 September 2016 - 04:16 PM

Very nice Jon, how would you say the noise compares to longer subs? I am guessing you backed the gain down to 200 because of noise?



#4 Jon Rista

Jon Rista

    ISS

  • *****
  • topic starter
  • Posts: 24,340
  • Joined: 10 Jan 2014

Posted 04 September 2016 - 04:22 PM

I actually backed gain down because of dynamic range. :p I was starting to clip the little helical finger in the bubble at gain 300. The FWC at that gain is not even 700e-, whereas at 200 the FWC is over 1900e-. Noise was not a problem at high gain...in fact, it was probably better. The difference in read noise between the two is only 0.17e- (~1.3e- @ 200, ~1.13e- @ 300)...so, a meaningless difference. 

 

If you had a fainter target and no immensely bright stars (or, did not mind clipping them a bit if you did), then I think gain 300 would be the better option. This kind of imaging is edging in on EMCCD territory. It's not as low read noise (with an EMCCD your read noise is going to be more like half an electron), however the signal is basically all photon shot noise once you calibrate with a dark. It's extremely clean, and you can see the difference between imaging at high gain with this camera, vs. imaging at lower gain settings with other cameras. The noise characteristic is so amazingly clean. 

 

The only thing I am disappointed about is my darn stars. :p I have perpetual tilt issues, and I'm about ready to give up trying to resolve them, as I don't think a resolution is possible. Not, at least, assuming I try to keep the filter wheel in play. I could get the Geoptik EF adapter, drop the filter wheel, and use the 1.25" filter holder that comes with the ASI1600 instead...but, that would certainly be far less convenient. 



#5 Jon Rista

Jon Rista

    ISS

  • *****
  • topic starter
  • Posts: 24,340
  • Joined: 10 Jan 2014

Posted 04 September 2016 - 04:36 PM

Here is an improved version of the SHO:

 

iL2kmiE.jpg

 

I tried to keep some green in the pillars around bubble, but that left an odd color cast everywhere. So I went with a more standard palette. 



#6 glend

glend

    Surveyor 1

  • *****
  • Posts: 1,955
  • Joined: 04 Feb 2014

Posted 04 September 2016 - 04:57 PM

Very nice Jon, thank for your continung R&D work with the ASI1600. 



#7 rockstarbill

rockstarbill

    Cosmos

  • *****
  • Posts: 7,564
  • Joined: 16 Jul 2013

Posted 04 September 2016 - 05:00 PM

Love it, Jon! *stalks his mailman for the Ha filter in route*



#8 calypsob

calypsob

    Fly Me to the Moon

  • *****
  • Posts: 5,830
  • Joined: 20 Apr 2013

Posted 04 September 2016 - 08:51 PM

Pretty spectacular imo john, curious, can you collect rgb in the red zone now that you have so much narrowband data? 



#9 Jon Rista

Jon Rista

    ISS

  • *****
  • topic starter
  • Posts: 24,340
  • Joined: 10 Jan 2014

Posted 04 September 2016 - 08:55 PM

Thanks Glen, Bill, Wes. 

 

Glen, welcome! :) I'm glad I'm actually able to do some real imaging now as well...until this point, it was all "R" for me. :p

 

Bill, hope your Ha gets there soon. It's fun getting high quality data despite the LP.

 

Wes, I don't think I'll be doing any LRGB in my red zone. I've never liked the look of light polluted color data. I do want to get back out to the dark site at some point, and when I do I'll be getting LRGB data on a few objects. Hopefully a few in one night, as this camera is significantly more efficient than my 5D III is. But, unlikely I'll do much LRGB from home. About all I expect is some experimental/testing/learning LRGB, so I have a better idea of what to expect when I finally to hit up the dark site. 



#10 FiremanDan

FiremanDan

    Aurora

  • *****
  • Posts: 4,869
  • Joined: 11 Apr 2014

Posted 04 September 2016 - 08:57 PM

This looks great on my phone!
Very interesting!

#11 rigel123

rigel123

    Hubble

  • *****
  • Posts: 18,145
  • Joined: 29 Jun 2009

Posted 04 September 2016 - 10:33 PM

Nice write up Jon and great results!



#12 Jon Rista

Jon Rista

    ISS

  • *****
  • topic starter
  • Posts: 24,340
  • Joined: 10 Jan 2014

Posted 04 September 2016 - 11:33 PM

Thanks Dan, Warren! 

 

I decided to process a little more. I extracted the luminance from the original combination, and used that as an L channel that I then combined with the original SHO version above, as well as a basic HOO version:

 

RFV3uzz.jpg

 

swKKsF4.jpg

 

I also fixed a remaining background gradient issue. 

 

Anyway, hopefully this shows what can be done with some high gain work. 



#13 Gary Imm

Gary Imm

    Mariner 2

  • -----
  • Posts: 245
  • Joined: 11 Apr 2016

Posted 07 September 2016 - 01:10 PM

Jon,

 

Excellent images and interesting post, as usual.

 

But I am still not clear on the advantage of imaging at a higher gain with this camera.  I have typically been imaging narrowband with this camera at unity gain with 5 minute exposures and I have been happy with the results so far.  Have you done any image comparisons of 300 gain vs unity gain, where the total integration time is the same?

 

Gary



#14 telfish

telfish

    3D Astro Solutions, custom 3D printing for Astrophotography

  • *****
  • Vendors
  • Posts: 2,309
  • Joined: 17 Nov 2010

Posted 07 September 2016 - 01:14 PM

Jon,

 

Excellent images and interesting post, as usual.

 

But I am still not clear on the advantage of imaging at a higher gain with this camera.  I have typically been imaging narrowband with this camera at unity gain with 5 minute exposures and I have been happy with the results so far.  Have you done any image comparisons of 300 gain vs unity gain, where the total integration time is the same?

 

Gary

 

 

The advantage for me is that I can get my NB subs down to 45 seconds and that means no guiding, So out goes the guide scope guide camera and associated software. 



#15 Jon Rista

Jon Rista

    ISS

  • *****
  • topic starter
  • Posts: 24,340
  • Joined: 10 Jan 2014

Posted 07 September 2016 - 01:31 PM

The main benefit is the short exposures. If you dial in a good enough PA, you could image unguided, which is kind of an unheard of thing for NB imaging. Another benefit is that at gain settings over unity you are better-sampling the electrons...0.485e-/ADU 12-bit for 200, 0.15e-/ADU for 300. Another benefit is that with shorter exposures, the amp glow is almost non-existent. So the impact to the noise profile across the image is so small that you can barely measure it, and you certainly cannot see it. The noise quality overall is just phenomenal...I haven't seen this quite before, not even with the ASI1600 at unity:

 

IySjpCu.jpg

 

The amount of tiny, faint stars I'm picking up is also amazing. I could NEVER do that with the 5D III and longer exposures, and haven't really seen it with 600s Gain 75 subs on the ASI1600. However, with these 90 second subs, there is considerably less blurring of those tiny stars, so they show up. They may only be a few pixels across (I was measuring FWHMs around 2.5", pixel scale is 1.3"), but they appear, they are very tight and small, they look wonderful. I've been chasing tight stars for years...it was so much harder with longer exposures on the 5D III (where I think 210-240 second subs @ ISO 1600 are about the ideal. Additionally, the low pass filter on the 5D III was probably blurring the crap out of things as well.) 

 

I haven't had a chance to do comparisons with other exposure lengths yet, although I do want to. I honestly don't know how things will compare, as I haven't had enough comparisons. The main comparisons I do have are my Gain 75 Pacman Ha data and the high gain Ha data I've acquired the last couple of nights. Statistically, stacking 36 high gain short exposures seems to be producing a much cleaner noise profile down into the faint signal areas than stacking 19 lower gain long exposures. I had some trouble pulling out the faint dust on Pacman with those 19 subs.

 

I think part of that may be due to the 12-bit precision and only 19 subs...it's just not enough to really restore a decent amount of precision...whereas 36 subs seems to be where you start benefiting from the statistical averaging and posterization does not seem to be a problem. Stacking only 16 high gain short exposure subs had similar problems to stacking 19 low gain long exposure subs as well...so I to think it is more of a statistical thing. It is almost effortless to get dozens of subs with the shorter high gain exposures, and certainly more effort, multi-night effort in my case, to get enough subs using longer exposures (and I never know if I'll have "another night" anymore)...and with this camera in particular due to the low bit depth, that is a very valuable thing. 


Edited by Jon Rista, 07 September 2016 - 01:34 PM.


#16 glend

glend

    Surveyor 1

  • *****
  • Posts: 1,955
  • Joined: 04 Feb 2014

Posted 07 September 2016 - 07:11 PM

Jon, I suspect that High Gain Narrowband is really only effective on faint targets, as bright objects will have you clipping all over the place - is that true in your experience?  Perhaps a lower gain layer mask could be used for bright nebula cores but bright stars would still just swamp the image would they not?



#17 Jon Rista

Jon Rista

    ISS

  • *****
  • topic starter
  • Posts: 24,340
  • Joined: 10 Jan 2014

Posted 07 September 2016 - 08:16 PM

I think it would depend on the exposure length. If you clip, back off the exposure a bit. Pacman is fairly bright. It might be clear tonight, so I'll see if I can get some high gain data on it and see how it goes. 



#18 roofkid

roofkid

    Viking 1

  • -----
  • Posts: 654
  • Joined: 13 Jan 2016

Posted 08 September 2016 - 05:57 AM

Hi everyone,

 

I didn't want to pollute the other image threads with a very specific question regarding your experiment. Basically I currently have an assumption and I want to put it out there to verify if I understood everything correctly. So in terms of narrowband imaging we assume that the sky background is almost 0 and we have "pure" signal + read noise + the "rest" of noise sources that should be lower in comparison.

 

Now my assumption. If I were to image under the same conditions with long vs short subs I would assume that I would loose all faint detail of the subject that does not get above the read noise floor in the exposure time. So if I were to image 60s subs and there's a region of my target which only statistically registeres 1 photon-electron per 60s, then I would not see that detail in a stack of 60s exposures while I would see it in a stack of 300s exposures. Would you concurr with that? Did I understand that correctly? Or am I mistaken on that?

 

What I'm trying to get at: Will I loose the faint stuff if I use lower subs or is it actually the other way around because I understood something wrong?

 

Thanks,

Sven



#19 keithlt

keithlt

    Surveyor 1

  • *****
  • Posts: 1,602
  • Joined: 21 Dec 2013

Posted 08 September 2016 - 09:00 AM

Great info Jon ! I have  some questions in dslr asa 800 was best for my 450D, kind of set it and forget it. is there a setting best for the ASI1600m? is gain and offset the same thing on the ASI1600m? would you find an advantage in combining different gains in an image?



#20 Jon Rista

Jon Rista

    ISS

  • *****
  • topic starter
  • Posts: 24,340
  • Joined: 10 Jan 2014

Posted 08 September 2016 - 10:35 AM

Hi everyone,

 

I didn't want to pollute the other image threads with a very specific question regarding your experiment. Basically I currently have an assumption and I want to put it out there to verify if I understood everything correctly. So in terms of narrowband imaging we assume that the sky background is almost 0 and we have "pure" signal + read noise + the "rest" of noise sources that should be lower in comparison.

 

Now my assumption. If I were to image under the same conditions with long vs short subs I would assume that I would loose all faint detail of the subject that does not get above the read noise floor in the exposure time. So if I were to image 60s subs and there's a region of my target which only statistically registeres 1 photon-electron per 60s, then I would not see that detail in a stack of 60s exposures while I would see it in a stack of 300s exposures. Would you concurr with that? Did I understand that correctly? Or am I mistaken on that?

 

What I'm trying to get at: Will I loose the faint stuff if I use lower subs or is it actually the other way around because I understood something wrong?

 

Thanks,

Sven

 

Signal adds linearly with stacking, while noise only grows in quadrature. In other words, signal grows fast, while noise grows slowly. So you should not have a problem pulling a faint sub-noise signal out of the noise if you calibrate and stack effectively.

 

So, even if you only got 1 photon per 60s sub, and 5 per 300s sub, you still get 5 total photons by stacking 5x60s subs. What changes is the SNR, because read noise compounds with sub count:

 

SNRlong = 5/SQRT(5 + 0.008*300 + 1.56^2) = 1.6:1

SNRshort = (1 * 5)/SQRT(5 * (1 + 0.008*60 + 1.1^1)) = 1.4:1

 

There is a small loss going with the short subs here. However, it doesn't take much to overcome that loss:

 

SNRshort = (1 * 7)/SQRT(7 * (1 + 0.008*60 + 1.1^1)) = 1.65:1

 

You may eventually hit some limits with stacking. If you do not totally remove any fixed pattern, and are not dithering effectively enough, any remnant pattern will eventually start to diminish the returns on stacking. I think it takes stacking quite a lot of data to reach that point, though. If you are going after something faint, and you simply cannot get it with short subs..then with a variable gain camera, the simple answer is to use a lower gain with longer subs. One of the benefits of a camera like the ASI1600 is it offers that flexibility.

 

Also keep in mind my original points.

 

A) High gain imaging is not necessarily about getting the deepest faintest details possible, it's about opening more NB imaging options

B) It can be useful in a pinch when you have limited clear sky time and want to create some full multi-channel NB in a single night

C) It is more about other benefits besides SNR: resolution, recovering full 16-bit precision, reducing sub loss risk, etc.



#21 Jon Rista

Jon Rista

    ISS

  • *****
  • topic starter
  • Posts: 24,340
  • Joined: 10 Jan 2014

Posted 08 September 2016 - 10:44 AM

Great info Jon ! I have  some questions in dslr asa 800 was best for my 450D, kind of set it and forget it. is there a setting best for the ASI1600m? is gain and offset the same thing on the ASI1600m? would you find an advantage in combining different gains in an image?

 

Gain is different than offset. Gain is the signal amplification, properly expressed in terms of e-/ADU. Offset is an offset, added to the pixel values to prevent clipping to black due to noise.

 

I generally don't see the value in combining data from different gain settings. When you change gain, you need calibration frames from a matching gain to calibrate, and if you mix and match gain settings, calibrating would be a real pain. On top of that, subs of different gain are going to be statistically different, so simple math becomes a less effective tool for gauging how well things should stack.

 

I don't think there necessarily is a "best" setting for the ASI1600. I think which setting is optimal for a given night, or even a given object, might depend a lot on your goals and your real-world limitations. If you have the time and option to work on a long-term project with longer subs and really go after FAINT details, Gain 75 might be better. However if you are just looking for a casual night, or have very limited sky time and are looking to make the most of the clear sky time you have, Gain 200 or 300 might be better, as you could acquire full multi-channel sets of data in significantly less time...even one night. You might not be able to go as deep as at a lower gain...but at least you can get something! If you are going after the highest resolution details possible, then imaging at a higher gain with shorter subs and being aggressive with culling will potentially deliver much higher resolution results than if you go with longer subs. 

 

This right here is one of the nice things about the ASI1600MM-Cool. It's got all of the flexibility of a DSLR (and more) as well as all of the flexibility of a planetary camera, packaged up into a simple, small, light weight, cooled, monochrome camera capable of all the same things as a CCD camera.



#22 telfish

telfish

    3D Astro Solutions, custom 3D printing for Astrophotography

  • *****
  • Vendors
  • Posts: 2,309
  • Joined: 17 Nov 2010

Posted 08 September 2016 - 01:33 PM

Some aspects of short sub high gain imaging that interest me are the ability to use slower scopes at a reasonable sub length. So you don't really need a super fast scope to image faint objects. That also opens up longer focal lengths as well.

 

The ability to do Narrow band imaging without the need to guide. As long as you can gain data with your sub length/ gain setting to get under your unguided imaging time limit you can avoid guiding altogether. For me that is a big advantage I avoid weight and 2 cables which in the winter makes a big difference.

 

I see no real downside right now, however it's early days and a lot more experimentation is needed to firm up the concept.



#23 Jon Rista

Jon Rista

    ISS

  • *****
  • topic starter
  • Posts: 24,340
  • Joined: 10 Jan 2014

Posted 08 September 2016 - 01:38 PM

Next clear night I get, I'll try and tackle CTB-1 with high gain. I guess if OU4 is in a good spot, I might try that first...it's OIII, so it would be the biggest challenge. My OIII data, even with twice the integration as Ha, is still a good deal noisier. I am honestly not sure why...I've got mostly high pressure sodium vapor lighting around me, so I'm not sure what might be polluting the OIII subs. But they definitely seem to be more light polluted than Ha. Oddly, same goes for SII. Same exposure lengths, the OIII and SII channels might even have better separation than Ha, but they are significantly noisier. I don't quite understand why that is...



#24 telfish

telfish

    3D Astro Solutions, custom 3D printing for Astrophotography

  • *****
  • Vendors
  • Posts: 2,309
  • Joined: 17 Nov 2010

Posted 08 September 2016 - 01:48 PM

Next clear night I get, I'll try and tackle CTB-1 with high gain. I guess if OU4 is in a good spot, I might try that first...it's OIII, so it would be the biggest challenge. My OIII data, even with twice the integration as Ha, is still a good deal noisier. I am honestly not sure why...I've got mostly high pressure sodium vapor lighting around me, so I'm not sure what might be polluting the OIII subs. But they definitely seem to be more light polluted than Ha. Oddly, same goes for SII. Same exposure lengths, the OIII and SII channels might even have better separation than Ha, but they are significantly noisier. I don't quite understand why that is...

 

 

Same here, my HA subs are pristine, Oiii acceptable and Sii very noisy. 



#25 dkeller_nc

dkeller_nc

    Surveyor 1

  • *****
  • Posts: 1,546
  • Joined: 10 Jul 2016

Posted 08 September 2016 - 09:47 PM

Jon - Thought I'd note that I'm giving this a try tonight based on your posts.  My Ha NB is considerably more modest than yours (12nm Optolong), and my target is different because of tree obstructions - it's the Iris Nebula.  But it'll be interesting to see what 10 - 20 90s subs in Ha will get me on this target, particularly with a 1/2 moon still in the sky.

 

By the way - I fiddled with the Sharp Cap Image Controls (specifically, the brightness setting).  I honestly can't see that it does anything, at least looking at live subs before processing.  I guessed and set it at 50, and took a few single subs while watching the histogram - no real difference that I can see.  I'm kind of wondering if SharpCap doesn't actually have the ability to apply an offset using the native ZWO driver (i.e., it might require using the ASCOM driver).




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics